Can I use specific models like o3 or gpt-5.2-codex?

Yes, the skill supports a --model flag that allows you to override the default and select specialized or reasoning-optimized models.

What are the prerequisites for using the Codex CLI skill?

You must have the OpenAI Codex CLI installed on your system and be authenticated with a valid OpenAI API key via the 'codex login' command.

Does it support multi-round feedback?

Yes, for plan reviews, the skill supports up to 3 rounds of iteration, relaying feedback until the plan is 'APPROVED' or the limit is reached.

How does this skill handle my codebase context?

For code reviews, it automatically reads git diffs. For plan reviews, you can provide file paths or snippets within the prompt, which Codex accesses in a read-only sandbox.

Codex CLI Integrator

Name: Codex CLI Integrator
Author: Awakehsh

byAwakehsh

•

Collaboration Tools

Integrates the OpenAI Codex CLI into Claude Code to perform high-reasoning code reviews and technical plan evaluations.

About

The Codex CLI skill bridges Claude's workflow with the powerful reasoning capabilities of the OpenAI Codex command-line tool. It enables developers to perform multi-round implementation plan reviews, analyze git diffs for uncommitted changes, and engage in complex problem-solving using specialized, code-optimized models. By leveraging external high-reasoning models, it provides an objective second opinion on architectural decisions and code quality, ensuring technical soundness before final deployment.

Key Features

Multi-round implementation plan review and approval tracking
Adjustable reasoning effort levels (medium, high, xhigh) based on task complexity
Customizable model selection supporting reasoning-heavy variants like o3 and gpt-5.2
Read-only sandbox integration for secure codebase context handling
1 GitHub stars
Automated code review for uncommitted changes and specific git branches

Use Cases

Validating a complex refactoring plan with high-reasoning models before implementation
Running automated logic and security reviews on uncommitted git changes
Brainstorming architectural solutions using code-optimized AI variants

About

Key Features

Multi-round implementation plan review and approval tracking
Adjustable reasoning effort levels (medium, high, xhigh) based on task complexity
Customizable model selection supporting reasoning-heavy variants like o3 and gpt-5.2
Read-only sandbox integration for secure codebase context handling
1 GitHub stars
Automated code review for uncommitted changes and specific git branches

Use Cases

Validating a complex refactoring plan with high-reasoning models before implementation
Running automated logic and security reviews on uncommitted git changes
Brainstorming architectural solutions using code-optimized AI variants